Distributed Resource Allocation Algorithm in Fog Networks with Energy Cooperation

Abstract

Fog networks with energy harvesting can extend the scope of services for users in time and space remarkably. However, the imbalance of the harvested renewable energy and the workloads of fog nodes (FNs) limits the performance of the whole fog computing system. This paper investigates on resource allocation in energy cooperation enabled fog computing networks, where FNs are connected with a smart grid and energy can be transferred between FNs. Firstly, an optimization problem is formulated to minimize the service latency of the fog computing system and is proved to be convex. After that, a cooperative graph-based distributed resource allocation (CGDRA) algorithm is proposed to solve the optimization problem by a distributed method. In our proposed CGDRA algorithm, an energy cooperative graph is constructed at first, then the FNs cooperate with their neighbors in energy by exchanging messages. Simulation results show that the CGDRA algorithm can balance the energy utilization ratios of different FNs and service latency of the fog network is reduced significantly. Moreover, the results obtained by our proposed CGDRA algorithm can approximate to the global optimal solutions with lower complexity compared with traditional centralized algorithm.